<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/149472>149472</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            [MLIR][Bindings][Transform] How to create foreach and include using MLIR python bindings
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            mlir
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          Dasor
      </td>
    </tr>
</table>

<pre>
    I'm trying to replicate the transform example listed in [this example ](https://github.com/llvm/llvm-project/blob/main/mlir/test/Integration/Dialect/Linalg/CPU/ArmSVE/matmul.mlir) (matmul SVE) but using the python bindings. I was able to replicate the whole` _tile_and_vectorize` named sequence but I'm not able to replicate the` __transform_main` as I can't find how to create both `transform.foreach` and `transform.include` How would something like that be done with the python bindings?. Here is the code I currently have:

```python
        def tileAndVectorize(func):
            sequence = transform.NamedSequenceOp(
 "__tile_and_vectorize",
 [transform.OperationType.get("func.func")],
                [],
 arg_attrs = [{"transform.consumed": UnitAttr.get()}],
 )
            with InsertionPoint(sequence.body):
                # Step 0: Get a handle to the matmul op
                matmuls = structured.MatchOp.match_op_names(
                    func,
 ["linalg.matmul"] 
                )
                
 # Step 1: Tile
                tiled = structured.TileUsingForOp(matmuls.result, sizes=[2,[4],1])

 # Step 2: Vectorize
                structured.VectorizeOp(tiled.results[0], [2, [4], 1])
                
                # Step 3: Lower vector.multi_reduction to vector.contract (+ some helpful patterns)
 with InsertionPoint(transform.ApplyPatternsOp(func).patterns):
 vector.ApplyVectorReductionToContractPatternsOp()
 vector.ApplyTransferPermutationPatternsOp()
 vector.ApplyLowerMaskedTransfersPatternsOp()
 vector.ApplySinkVectorPatternsOp()
 vector.ApplyLowerMaskedTransfersPatternsOp()
 vector.ApplyLowerContractionPatternsOp(lowering_strategy=vector.VectorContractLowering.OuterProduct)
 vector.ApplyLowerOuterProductPatternsOp()
 
                transform.YieldOp([])

        def transform_main():
            sequence = transform.NamedSequenceOp(
 "__transform_main",
                [transform.AnyOpType.get()],
 [],
                arg_attrs = [{"transform.consumed": UnitAttr.get()}],
            )
            
            with InsertionPoint(sequence.body):
                # get all the functions
 funcs = structured.MatchOp.match_op_names(
 sequence.bodyTarget,
                    ["func.func"]
 )
                ## MISSING: how to use transform.foreach + transform.include ???

                transform.YieldOp([])
```

as said the `tileAndVectorize` already works, but I don't know what to put in the `transform_main`. Any help is appreciated, thanks in advance.
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJy0V2-P2rgT_jTmjfWLgkMWeMELWH60SLvdVXdb6V4hJx6Ib42ds53l6Kc_jZ3wr-lVlXoWUgTjmXn8zHh4wp2TOw0wI_mC5MsBb3xl7GzJnbGDwojjbE3YeE-9PUq9o95QC7WSJfdAfQXUW67d1tg9hb_5vlZAlXQeBJWaknzhK-lOFpIvCZtU3teOZHPCVoStdtJXTZGUZk_YSqn37vG_2po_ofSErQplCsJWey41PpS0hK08OLSttYed5V4atC0lV9HnQWqudoSt7p-_ELaa2_3L1_-HIH7fqCQGmVLCJvEXGsxTWjSeNi6ctAJaH31lNC2kFlLvXELX9MAd5YWC75k4VEYBuUvpxksFG67F5h1Kb6z8Fn7WfA-COvirAV1CyBSp1cb3hwzBNieGN4GBu5RyR9e05JqwsadbqQWtzAG9SwvoWhhfUXKXnjyTrbHAyyo4a3Ftk7pUjQjJPpoDPZhGCerMHnyFPCj5hmC4pwVQYTTQg_RVHz0kWyX0I1ig0gV7aQQg0sZa0F4dacXfASufhs9dGj8xDEnntF0CthQ5nGvx9cQgm2wbXRI2jQHoxTpxSrLluSGTT0j4S2t7qgmboB9hbNNXIcYIuw8b8sU5xlMNsbtejzUkO_AYhTGEkkQ8DCFhY9_foMIVL1Vr43a34d5bF3CiabwgjJ2TlUa7Zg8Cg2Zz-kVLP_fentJOyXh5DoffrzOGwqy1A4uIn43U6Nexk-Bt7uUvIGUZffFQ0xRTfwBPOa24FrEtsZrtRTF1j3e0xYM5b5vSNxZE8sh9WT3VyR6fG1Nv8A64thA9KzLaVYEwpsI1TmJ4pCVf0l7wt1SEXwNJ7bGGeKxXqaBnI3aDuMWOe7_gJFgZG5qnPWNiwTXKE3ZPnfwGjmRLki9C8-SLUSzPMDymsc_PGEJVzx39PZCL9KdtIXdA2GZ2JF-kMQ9tM9NzanrO3cfHj6qeIbQHcwBL45VI9o3ycmNBNCV2E3ZBaymN9paXnoamXIRZQStQ9bZRtObeg9WuhdDfkueWn9e1Oj63PuGo7S1PLgPFjm3TB5dIz-cO3au5b0FdxWpBXDq-htRgn8HuGx-u9s9cAi2P3L2B6Lzdz3xepH6LGP-L6MGnO_HtCRQapd5tnLfcw-5IsmXrHBF1jg_txuSp8WCfrUEyf5Ttck8fvL5rdaryHxKUiLvbgdjdjXaFiX_9Pxcj_45ZfxOX_XBUX7SlPj7VVyP_YsZfDfWb9Ztn_L_OuN88_Xc485UKsx4vIUZxuBe__Npov0r6ym04XB9dLfG3_6j5sv8ProWKaB_XLy_rTx-Qw1b8NO5CjXaKh-KA-k7rUJKt2k_aR8dPO7eTLtGdO-q4FIE5lFa30gVVl7LAxZEejH1zOKaD_EM9FUTcm0blhSLLG1o3HtVzF-1W_iV0ro9h3qLO4nVtoZTcY0vdo1DTbw7duXjnWIKBmGVimk35AGbDcc4mbJxOs0E1G4nt6K4U2WSbFRMxHGdDUfKpGLI7kZeClQM5YynL0_Fwkk7YKM-SUVGwUQk5ZHfbsoAhGaWw51IlKNgTY3cD6VwDs-FoOhqzgeIFKDeL9Y2KG0s7sLMg8Itm58goxdcFd47gpVfhZeTxYf0ZCc8Xi05fhm-vHR-oBT5e6d6u5ihxu0pHMY_BbuXqoLFq9ssvI-GEjrBVe8j3GfsnAAD__yn8IK8">